TABLE 1.

Description of commonly used filtering approaches in the analysis of RADseq data (“filter”), the reason for their usage (“usage”), and how they impact population genomic inference (“impact”)

Filter	Usage	Impact	Reference
Hardy–Weinberg equilibrium (HWE)	Removes loci under selection Removes library and sequencing artefacts	Unknown	Gruber et al. (2018), Sethuraman et al. (2019), Waples (2015)
Linkage within loci	Mitigates effects of nonindependence of single nucleotide polymorphisms (SNPs) by removing physically linked SNPs.	Reduces false signals of population structure Necessary for STRUCTURE (if LD correction is not used)	O'Leary et al. (2018)
Locus level diversity	Loci with high SNP density (i.e., many SNPs within a locus) may be the result of polyploidy	Can remove putative paralogous loci	Hohenlohe et al. (2011), Mastretta‐Yanes et al. (2015)
Minor allele frequency (MAF)/count (MAC)	Identification of genotyping errors	Can remove informative loci if not applied carefully MAF will affect loci differently based on missingness Removes genotyping errors	Linck and Battey (2019), O'Leary et al. (2018)
Variant call rate	Ensures SNP panel is well represented across individuals	Can dramatically reduce number of loci Helps ensure samples are comparable	O'Leary et al. (2018)